I seriously doubt it's possible to alter audio tracks without separating them from the video first. You have to demux the audio stream, modify it, then remux it back with the video. Even programs like transcode always do this when they do their work. But I suppose in this case you just want to know how to get transcode to pass the video stream through unchanged.
explains how to pass through video streams while processing the audio. I'm not sure if the 'raw' module will work with all video types though.
What I've always done in the past myself is write up a bash script to automate manually running a demux-reencode-remux process, usually with the underlying tc tools included rather than with transcode proper. In my experience, working with the streams separately often leads to less problems than trying to get the program to do everything at once.